|Publication number||US7856281 B2|
|Application number||US 12/267,039|
|Publication date||21 Dec 2010|
|Filing date||7 Nov 2008|
|Priority date||30 Sep 2005|
|Also published as||CN1940780A, CN1940780B, CN101807048A, CN101807048B, DE102006045429A1, US7451004, US20070078529, US20090143872|
|Publication number||12267039, 267039, US 7856281 B2, US 7856281B2, US-B2-7856281, US7856281 B2, US7856281B2|
|Inventors||Dirk Thiele, Wilhelm K. Wojsznis|
|Original Assignee||Fisher-Rosemount Systems, Inc.|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (146), Non-Patent Citations (53), Referenced by (20), Classifications (11), Legal Events (4)|
|External Links: USPTO, USPTO Assignment, Espacenet|
This application is a divisional of U.S. patent application Ser. No. 11/240,705 filed Sep. 30, 2005, entitled “On-Line Adaptive Model Predictive Control in a Process Control System,” the entire disclosure of which is hereby incorporated by reference herein.
The present invention relates generally to process control systems and, more particularly, to the creation and use of an on-line adaptive model predictive controller or other model predictive control type controller within a process control system.
Process control systems, such as distributed or scalable process control systems like those used in chemical, petroleum or other processes, typically include one or more process controllers communicatively coupled to each other, to at least one host or operator workstation and to one or more field devices via analog, digital or combined analog/digital buses. The field devices, which may be, for example valves, valve positioners, switches and transmitters (e.g., temperature, pressure and flow rate sensors), perform functions within the process such as opening or closing valves and measuring process parameters. The process controller receives signals indicative of process measurements made by the field devices and/or other information pertaining to the field devices, uses this information to implement a control routine and then generates control signals which are sent over the buses to the field devices to control the operation of the process. Information from the field devices and the controller is typically made available to one or more applications executed by the operator workstation to enable an operator to perform any desired function with respect to the process, such as viewing the current state of the process, modifying the operation of the process, etc.
Process controllers are typically programmed to execute different algorithms, sub-routines or control loops (which are all control routines) for each of a number of different loops defined for, or contained within a process, such as flow control loops, temperature control loops, pressure control loops, etc. Generally speaking, each such control loop includes one or more input blocks, such as an analog input (AI) function block, a single-output control block, such as a proportional-integral-derivative (PID) or a fuzzy logic control function block, and a single output block, such as an analog output (AO) function block. These control loops typically perform single-input/single-output control because the control block creates a single control output used to control a single process input, such as a valve position, etc. However, in certain cases, the use of a number of independently operating, single-input/single-output control loops is not very effective because the process variables being controlled are affected by more than a single process input and, in fact, each process input may affect the state of many process outputs. An example of this might occur in, for example, a process having a tank being filled by two input lines, and being emptied by a single output line, each line being controlled by a different valve, and in which the temperature, pressure and throughput of the tank are being controlled to be at or near desired (set point) values. As indicated above, the control of the throughput, the temperature and the pressure of the tank may be performed using a separate throughput control loop, a separate temperature control loop and a separate pressure control loop. However, in this situation, the operation of the temperature control loop in changing the setting of one of the input valves to control the temperature within the tank may cause the pressure within the tank to increase, which, for example, causes the pressure loop to open the outlet valve to decrease the pressure. This action may then cause the throughput control loop to close one of the input valves, thereby affecting the temperature and causing the temperature control loop to take some other action. As will be understood in this example, the single-input/single-output control loops cause the process outputs (in this case, throughput, temperature and pressure) to behave in an unacceptable manner wherein the outputs oscillate without ever reaching a steady state condition.
Model predictive control (MPC) or other types of advanced control which use dynamic matrix control (DMC) techniques have been used to perform process control in situations in which changes to a particular controlled process variable affects more than one process variable or output. Since the late 1970s, many successful implementations of model predictive control have been reported and MPC has become the primary form of advanced multivariable control in the process industry. Still further, MPC control has been implemented within distributed control systems as distributed control system layered software. U.S. Pat. Nos. 4,616,308 and 4,349,869 generally describe MPC controllers that can be used within a process control system.
Generally speaking, MPC is a multiple-input/multiple output control strategy in which the effects of changing each of a number of process inputs (i.e., manipulated variables) on each of a number of process outputs (i.e., controlled variables) is measured and these measured responses are then used to create a control matrix for use in controlling the process. The control matrix includes a process model (which generally defines the dynamic operation of the process) which must be inverted mathematically and then used in or as a multiple-input/multiple-output controller to control the process outputs based on changes made to the process inputs. In some cases, the process model is represented as a process output response curve (typically a step response curve) for each of the process inputs and these curves may be created based on a series of, for example, pseudo-random step changes delivered to each of the process inputs. These response curves can be used to model the process in known manners. Model predictive control is known in the art and, as a result, the specifics thereof will not be described herein. However, MPC is described generally in Qin, S. Joe and Thomas A. Badgwell, “An Overview of Industrial Model Predictive Control Technology,” AIChE Conference, 1996.
While MPC is useful in many process control situations, MPC applied in the industry predominantly uses a dynamic matrix control (DMC) technique that requires generation of a typically complex process model (typically mathematically represented as a recursive algorithm or a matrix calculation) and then subsequent inversion of that model to create the control matrix used in the MPC controller. As a result, generating the MPC controller is computationally expensive. Still further, in order to develop an accurate process model for use in creating an MPC controller, it has traditionally been necessary to disturb or upset the process with known control signals, e.g., step signals, for each of the control inputs and determine the reaction of each of the process variables or controlled variables to the known changes in the control inputs. One way of implementing this procedure in a distributed process control system is described in more detail in U.S. Pat. No. 6,721,609, entitled “Integrated Optimal Model Predictive Control in a Process Control System”, the disclosure of which is hereby incorporated herein.
While this process disturbance technique generally produces a highly accurate process model, the disturbance procedure takes time and upsets the normal operation of the process and is thus difficult if not practically impossible to perform when a process is running on line. Instead, this process disturbance technique typically needs to be implemented on the process when the process is not operating to create an actual product, e.g., during the initial configuration of the process or the MPC controller. Of course, this constraint severely limits the times during which a process model can be determined. In any event, this technique is not suitable for use in an adaptive MPC controller (i.e., one in which the MPC control matrix is changed during the on-line operation of the process) as it requires process upset during each adaptive cycle. Furthermore, this method is computationally expensive for MPC controllers of any significant size (i.e., those having multiple inputs and multiple outputs) as the control matrix must be inverted and applied after determining a new process model. This computational load makes it difficult to implement adaptive MPC in distributed process controllers, which are typically limited in the amount of additional computational load that can be performed in conjunction with performing on-line process control activities.
However, in many situations, it is desirable to adapt the MPC controller during operation of the process to account for process model mismatch. In particular, when implementing MPC, the process model determined at the configuration stage only reflects the process at the time that the process model was created. Any subsequent changes in the process, which changes typically occur naturally during the course of running a process, will not be reflected in the process model used by the MPC controller and may therefore lead to model mismatch and non-optimal control by the MPC controller. MPC controllers are most susceptible or sensitive to modeling errors in the dead time of the process loop. In many control situations, it is desirable and sometimes necessary to compensate for this model mismatch.
In the past, one method used to compensate for model mismatch in a DMC or other MPC controller was to periodically create a new process model and generate a new control model and controller using the process upset or disturbance technique. However, as described above, this procedure could only be performed infrequently and computation has to be performed off-line due to the need to perform process upset to determine a new process model and due to the amount of computations needed to be performed during the controller matrix generation process. Another manner of compensating for model mismatch for non-linear processes is described in U.S. patent application Ser. No. 10/454,937, entitled “Multiple-Input/Multiple-Output Control Blocks with Non-Linear Predictive Capabilities,” which uses a non-linear process model in conjunction with an MPC controller generated using a process upset technique. Generally speaking, this technique compares predicted process changes developed using a non-linear process model and predicted process changes developed from the MPC process model to create error signals indicative of the model mismatch, and then uses these error signals to compensate for non-linear characteristics of the process which were not accounted for or modeled when generating the MPC controller. However, this technique relies on the use of a complex non-linear process model to accurately reflect the process operation and to create appropriate compensation signals for the MPC controller. Additionally, in many cases, there may still be model mismatch between the non-linear process model and the actual process, which can lead to poorer control performance. Still further, this technique is not adaptive, as neither the non-linear process model nor the MPC process model is changed during the on-line operation of the controller.
Partially as a result of the difficulty in creating and implementing an effective adaptive MPC controller or other type of adaptive DMC controller, the art of process control uses adaptive PID controllers in many process situations in which the process model changes frequently during operation of the process. While adaptive PID controllers are well known and are applied to adapt during the operation of a process, these PID controllers, to operate satisfactorily, must be detuned or tuned very conservatively in dead time dominant processes, which leads to poor performance. Still further, PID controllers are highly susceptible to poor control performance when there is a mismatch between the controller reset and the actual process time constant, especially when the process dynamics change frequently. Unfortunately, determining the process time constant (which is directly related to the process time to steady state) is the most uncertain parameter developed when using known process model identification methods. Thus, PID controllers are not necessarily the best choice when controlling a dead time dominant process, especially when the time to steady state of the process changes frequently.
A method of creating and using an adaptive DMC or other MPC controller includes using a model switching technique to periodically determine a process model, for example, a parameterized process model, for a process loop on-line during operation of the process, without having to artificially excite the process. The method then uses the process model to generate an MPC control model and creates an MPC controller algorithm on-line, i.e., while the process in executing normally. This technique, which is generally applicable to single-loop MPC controllers and is particularly useful in MPC controllers with a control horizon of one or two, enables an MPC controller to be adapted on-line, i.e., during the normal operation of the process, to change the process model on which the MPC controller is based and to thereby account for changes in the process over time. Such an adaptive MPC controller is not computationally expensive and can therefore be easily implemented within a distributed controller of a process control system while providing the same or in some cases better control than a PID controller. More particularly, it has been found that an adaptive, single-loop MPC controller with a small control horizon, e.g., one or two, can provide better control than a PID controller in dead time dominant loops, and especially in cases in which the process loops are subject to process model mismatch due to changes in the process dynamics. Additionally, an MPC controller can accommodate easily more than one feedforward input usually not available in PID controllers.
Referring now to
The field devices 15-22 may be any types of devices, such as sensors, valves, transmitters, positioners, etc. while the I/O cards 26 and 28 may be any types of I/O devices conforming to any desired communication or controller protocol. In the embodiment illustrated in
The controller 11, which may be one of many distributed controllers within the plant 10 having at least one processor therein, implements or oversees one or more process control routines, which may include control loops, stored therein or otherwise associated therewith. The controller 11 also communicates with the devices 15-22, the host computers 13 and the data historian 12 to control a process in any desired manner. It should be noted that any control routines or elements described herein may have parts thereof implemented or executed by different controllers or other devices if so desired. Likewise, the control routines or elements described herein to be implemented within the process control system 10 may take any form, including software, firmware, hardware, etc. For the purpose of this discussion, a process control element can be any part or portion of a process control system including, for example, a routine, a block or a module stored on any computer readable medium. Control routines, which may be modules or any part of a control procedure such as a subroutine, parts of a subroutine (such as lines of code), etc. may be implemented in any desired software format, such as using ladder logic, sequential function charts, function block diagrams, object oriented programming or any other software programming language or design paradigm. Likewise, the control routines may be hard-coded into, for example, one or more EPROMs, EEPROMs, application specific integrated circuits (ASICs), or any other hardware or firmware elements. Still further, the control routines may be designed using any design tools, including graphical design tools or any other type of software, hardware, or firmware programming or design tools. As a result, the controller 11 may be configured to implement a control strategy or control routine in any desired manner.
In one embodiment, the controller 11 implements a control strategy using what are commonly referred to as function blocks, wherein each function block is a part or object of an overall control routine and operates in conjunction with other function blocks (via communications called links) to implement various process control loops within the process control system 10. Function blocks typically perform one of an input function, such as that associated with a transmitter, a sensor or other process parameter measurement device, a control function, such as that associated with a control routine that performs PID, fuzzy logic, etc. control, or an output function which controls the operation of some device, such as a valve, to perform some physical function within the process control system 10. Of course hybrid and other types of function blocks exist. Function blocks may be stored in and executed by the controller 11, which is typically the case when these function blocks are used for, or are associated with standard 4-20 ma devices and some types of smart field devices such as HART devices, or may be stored in and implemented by the field devices themselves, which may be the case with Fieldbus devices. While the description of the control system is provided herein using a function block control strategy which uses an object oriented programming paradigm, the control strategy or control loops or modules could also be implemented or designed using other conventions, such as ladder logic, sequential function charts, etc. or using any other desired programming language or paradigm.
As illustrated by the expanded block 30 of
As will be described in more detail, the adaptive MPC control block 38 is a control block that monitors the process and that re-calculates a process model, such as a parameterized process model, for the process (or at least the portion or loop of the process being controlled by the control block 38) on-line, during operation of the process. The adaptive MPC control block 38 then uses the new process model, whenever determined, to recalculate the MPC control model and the MPC control algorithm used within the control block 38, to thereby adapt the MPC controller within the MPC control block 38 to better match or to control based on the newly calculated process model. This adaptive MPC control occurs without the need to artificially upset the process to determine the new process model and without having to take the process off-line to calculate and install the new MPC controller model and algorithm within the MPC controller. As will be understood, the process model can be re-defined and the MPC controller can be regenerated at various times during operation of the process to reduce or eliminate model mismatch between the MPC controller and the process due to changes in the process over time.
As noted above, while the adaptive control block 38 will be described herein as including a model predictive control (MPC) block, the control block 38 could implement other DMC type control techniques instead using the same principles described herein. Still further, it will be understood that the function blocks illustrated in
As illustrated in
The MPC controller block 54 receives, as inputs, a measured controlled variable CV (as measured within the process 50), a disturbance variable DV, a set point value or vector SP defining the desired target vector for the controlled variable CV, and the manipulated variable MV produced by the controller block. As is known, the disturbance variable DV represents measured or predicted change (e.g., a disturbance) in the process 50 and is illustrated as being provided to the process 50 at the same time as this value is provided to the controller block 54. Generally speaking, the disturbance variable DV represents the input to the feedforward path of the MPC controller 54 while the controlled variable CV represents the input to the feedback back path of the MPC controller 54.
As is typical in an MPC controller, the controlled variable CV and the disturbance variable DV are provided, along with the manipulated variable MV produced by the MPC controller 54, to the input of a controlled variable process model 70 (also referred to as a controlled variable prediction unit). The controlled variable prediction unit 70 uses a process model (referred to a control model) stored therein to predict the future values of the controlled variable CV based on the current and/or predicted future values of the manipulated variable MV and the disturbance variable DV. (Typically, a separate control model is used for each of these input variables). The controlled variable prediction unit 70 produces an output 72 representing a previously calculated prediction for the controlled variable CV for the current time and a vector summer 74 subtracts the predicted value of the controlled variable CV for the current time from the actual measured value of the controlled variable CV to produce an error or prediction correction vector on an input 76.
Generally speaking, the controlled variable prediction unit 70 uses a step response matrix (which in this case may be mathematically developed from a process model) to predict a future value for the controlled variable CV at each of the times over the prediction horizon, based on the disturbance and manipulated variables DV and MV and the error signal provided to other inputs of the controlled variable prediction unit 70. The output of the unit 70 is illustrated as a predicted CV vector. A set point prediction unit 80 provides a target vector for the controlled variable CV based on the set point SP provided thereto from any desired source, such as an optimizer, a user, a control operator, etc. In one embodiment, the set point prediction unit 80 may use the set point SP along with a pre-established change or filter vector which defines the manner in which controlled variable CV is to be driven to its set point value over time (i.e., defining the robustness and speed of the controller). The set point prediction unit 80 produces a dynamic control target vector (called a predicted CV target vector) for the controlled variable CV defining the changes in the set point value for the controlled variable CV over the time period defined by the prediction horizon. A summer 84, which may be a vector summer, then subtracts the predicted CV vector from the dynamic control target vector to define a future error vector for the controlled variable CV. The future error vector for the controlled variable CV is then provided to an MPC algorithm block 85 which operates to select the manipulated variable MV steps (for each of the time periods up to the control horizon) that minimize, for example, the least squared error, over the control horizon. Of course, the MPC algorithm block 85 may use a control matrix or other algorithm developed from relationships between the controlled variable CV and the disturbance variable DV input to the MPC controller 52 and the manipulated variable output by the MPC controller block 54. As is generally known, the MPC algorithm block 85 tries to minimize controlled variable CV error with minimal manipulated variable MV moves over the control horizon, while keeping the controlled variable CV within operational constraints and while achieving a steady state value for the manipulated and controlled variables MV and CV within a limited amount of time.
Generally speaking, the MPC controller block 54 operates once during each controller scan to produce a single-loop MPC control signal as the controlled variable CV to be used in controlling the process 50 based on the MPC control model and the control algorithm stored within the blocks 70 and 85, respectively. However, as noted above, the dynamics of the process 50 typically change over time, which can lead to a mismatch between the actual operation of the process 50 and the model of the process 50 being used within the MPC controller block 54.
To compensate for this problem, the adaptive model generator 52 operates to re-determine or update a process model representing the process 50 and, in particular, representing the loop of the process 50 being controlled by the MPC control block 54. The adaptive model generator may determine the same or separate process models for the feedback and the feedforward path of the MPC controller 54, if desired. The adaptive model generator 52 then uses the updated process model to determine a new MPC control model for use in the block 70 as well as a new MPC control algorithm for use in the block 85 to thereby enable the MPC controller 54 to operate based on a process model that more accurately reflects the current operation of the process 50. This updating process adapts the MPC controller 54 to the process 50 on-line during operation of the process 50 to thereby eliminate or reduce model mismatch and provide better control.
Generally speaking, the adaptive model generator 52 includes a process model estimator 90 which operates to determine or recalculate, on-line and while the process 50 is running, new models for the process 50 and, in particular, for the specific loop of the process 50 which is being controlled by the MPC controller block 54. The output of the process model estimator 90 is a process model, such a parameterized process model which defines the operation of the process 50 according to a set of parameters. The most common type of parameterized process model is a first order plus dead time process model which includes parameters for the process response time, the process gain and the process dead time, although other parameterized process models could be used instead. A method for defining or estimating process models from process variables for use in an adaptive PID controller is described in U.S. Pat. No. 6,577,908 entitled “Adaptive Feedback/Feedforward PID Controller,” and U.S. Publication No. 2003/0195641, entitled “State Based Adaptive Feedback Feedforward PID Controller,” the disclosures of both of which are hereby expressly incorporated by reference herein.
Generally speaking, the process model estimator 90 regularly collects data indicative of the controlled variable CV and one or more of the manipulated variable MV, the disturbance variable DV, and the set point SP, and possibly other variables if desired during normal operation of the process. The process model estimator 90 then periodically reviews or analyzes this data (or enables a user to do so via the user interface routine 44 of
After determining a new process model, the process model estimator 90 provides the process model (e.g., the calculated process gain K, dead time DT and time constant Tc) to an MPC model calculation unit 92 as well as to a penalty on move and set point target vector filter unit 94. The MPC model calculation unit 92 uses the new process model to calculate a typical MPC control model to be used in the controlled variable prediction unit 70. This MPC control model is generally in the form of a transfer function represented as a response curve defining the response of the controlled variable CV to a step change in the manipulated variable MV (or the disturbance variable for the feedforward path) over the time to the prediction horizon. This model is generally easy to mathematically calculate as the series of values of the controlled variable CV (one for each of the scan times up until the prediction horizon) which would result from a process defined perfectly by the parameterized process model developed by the process model estimator 90 in response to a step change in the manipulated variable MV.
After determining the MPC control model, the MPC model calculation unit 92 provides this model to an MPC algorithm calculation unit 96 and stores this model for future use or downloading into the controlled variable prediction unit 70 of the MPC controller 54 when updating the MPC controller 54. At essentially the same time, the penalty on move and set point target vector filter unit 94 calculates or otherwise determines a penalty on move and a set point trajectory or filter coefficients to be used in the MPC controller 54. In one embodiment, as described in more detail hereinafter, the set point target vector filter trajectory and the penalty on move, which are control tuning variables used in the SP prediction unit 80 and the MPC algorithm block 85 respectively, may be automatically calculated. For example, the penalty on move may be automatically based on the process time to steady state Tss which, in turn, is determined from response time of the process model developed by the process model estimator 90. In another embodiment, the penalty on move and the set point target vector filter trajectory may be input or specified by a user, such as a control operator.
An MPC algorithm calculation unit 96 uses the MPC control model (and typically inverts this model) as well as the penalty on move value to determine an appropriate control algorithm for use by the MPC algorithm block 85 based on the newly determined process model. Thereafter, at an appropriate time, such as when the process 50 is in a steady state or a quasi-steady state condition, or when instructed to by a user via a user interface, the adaptive model generator 52 updates the MPC controller block 54 by downloading the new MPC control model to the controlled variable prediction unit 70, the new SP trajectory or filter coefficients to the SP prediction block 80 (if changed), and the new MPC control algorithm to the MPC algorithm block 85.
In this manner, when the process model estimator 90 determines or detects a process model for the process 50 which differs in some significant manner from the process model that was used to configure the MPC controller 54, the MPC model calculation unit 92 and the MPC algorithm calculator block 96 may calculate new MPC controller parameters, models, and algorithms based on that model and then download these new MPC controller elements to the MPC controller 54.
Significantly, it has been determined that a single-loop MPC or other DMC-type controller having a control horizon of one or two, and possibly higher, can be adapted as described above without being computationally expensive, and therefore, that such an adaptive controller can be run or executed in a distributed or other process controller on-line while the process is running, thus providing a truly on-line adaptive MPC (or other DMC type) controller. In particular, it has been determined that the MPC control algorithm (as used in the block 85 of
Still further, as noted above, it has been determined that single-loop MPC controllers including MPC with a small control horizon (e.g., one or two and possibly up to five) provide better control performance than PID controllers in many situations, such as in dead time dominant processes, and in processes in which the time to steady state changes over time.
More particularly, a block 104 which may be implemented by the process model estimation unit 90, collects process input/output data and a block 106 determines whether enough data has been collected over a time period in which the process has undergone a change sufficient enough to calculate a new process model. The block 106 may also be responsive to user commands to generate a process model from a selected set of process data. If the collected data is not such that the process model estimation unit 90 can calculate a new process model or the user has not instructed the routine 100 to develop a new process model, the block 104 continues to collect process input/output data. On the other hand, if enough process data has been collected over a time period in which the process has undergone a change or disturbance significant enough to calculate a new process model, or if a user has initiated a process model calculation, a block 108 calculates a new process model using, for example, the technique described in U.S. Pat. No. 6,577,908 and/or in U.S. Publication No. 2003/0195641 and provides the new process model to a second section 110 of the routine 100.
The section 110 of the routine 100 determines a new MPC model and algorithm from the determined process model. For the sake of this description, the process model determined by the block 108 will be assumed to be a parameterized first order plus dead time process model including parameters defining a process gain, a process dead time and a process time constant for both the feedback path and the feedforward path. However, it will be understood that the process model could include parameters only for the feedback path, could be a first order process model of a different type or nature or could be any type of process model other than a first order process model. Thus, generally speaking, other types or forms of process models could be used instead.
As illustrated in
Similar to an adaptive PID control technique, the time to steady state Tss is updated after every successful model identification based on the process model and in particular, based on the time constant of the process model. However, unlike a PID controller which can run at any scan rate (as long as the scan rate is a couple of times faster than the process response time), the adaptive MPC controller must run at a scan rate that keeps the time to steady state Tss inside of the prediction horizon used by the MPC controller. In order to alter or change the scan rate of the MPC controller as part of the adaptive process and thereby provide better control, it has been determined that the execution time and or the prediction horizon used by the MPC controller (which are typically fixed in MPC controllers) can be changed, to thereby provide for better MPC controller operation.
To assure that the new scan rate is selected appropriately, the block 112 first determines the process time to steady state Tss from the new process model in any known manner and, in particular, based on the time constant determined in the process model for, for example, the feedback path of the MPC controller 54. The time to steady state Tss may be determined or expressed in actual time (e.g., in minutes, seconds, etc.) or may be expressed in the number of executions cycles needed by the controller during the time to steady state based on fixed controller execution time per execution cycle.
Next, the execution time can be calculated as the time to steady state Tss divided by the maximum allowable prediction horizon, which may be set by a user or by a configuration engineer during configuration of the adaptive MPC controller. This calculation can be expressed as:
Exec_Time=Trunc[T ss /PH max]
In an embodiment described, a maximum prediction horizon PHmax of 120 is used, which means that up 120 predictions may be calculated by the controlled variable prediction unit 70 during each scan. However, in many cases, this calculation leads to a reminder, which causes more or less imprecision in the operation of the MPC controller. To avoid division with a remainder and thereby to reduce floating point error and jitter in the controller operations, the prediction horizon may be allowed to vary between a minimum value PHmin and the maximum value PHmax and may be chosen so that the product of the selected prediction horizon PH and the execution time equals the time to steady state Tss exactly. The selected prediction horizon is then used as the prediction horizon within all of the internal loops within the MPC controller (i.e., within all of the blocks 70, 74, 80, 84, and 85 the controller block 54 of
The following Table 1 illustrates examples of various combinations of execution times Exec_Time and selected prediction horizons PH which may be advantageously used within the MPC controller 54 based on a particular time to steady state Tss of the new process model. As will be understood, the time to steady state Tss of this table (which is the known variable from which the execution time and the prediction horizon are determined) is varied between 10 and 12800 seconds in this case, while the execution time varies between 0.1 to 213 ms, resulting in a selected prediction horizon PH somewhere between 60 and 120 in all cases. While not necessary, the prediction horizon is set to be 60, i.e., the minimum allowable value, for each time to steady state value greater than 3200 because a prediction horizon greater than 60 is not necessary for such a large time to steady state. In any event, if desired the results of Table 1 or a similar such pre-calculated table relating a desirable combination of a prediction horizon and an execution time to different determined process time to steady states Tss may be stored in the adaptive model generator 52 to be used to determine the appropriate prediction horizon to use in the MPC controller based on the determined process model time to steady state Tss. Of courses any other desired manner of generating combinations of the prediction horizon and the execution time based on a determined process time to steady state may be used instead.
In any event, as will be seen, the adaptive MPC controller block 38 is actually able to change or alter its prediction horizon and execution time during the adaptation process in order to assure that the controller scan rate is set so as to keep the time to steady state Tss within the prediction horizon. If desired, the block 112 may compare the execution time to the scan rate of the controller to determine if the selected or calculated execution time is smaller than the configured scan rate of the controller. If this condition is true, the block 112 may issue a warning message to a user via, for example, the operator interface application 44, indicating that the controller block scan rate should be increased to enable the proper operation of the MPC controller.
Next, a block 114, which may be implemented in the MPC model calculation block 92 of
Next, a block 116 calculates or determines the penalty on move and the penalty on error to be used in the MPC algorithm and, if desired, the set point filter constant or trajectory to be used in the MPC controller to determine the response and robustness of the MPC controller 54. These parameters are, in essence, tuning parameters for the MPC controller. If desired, the penalty on error, the SP filter and the penalty on move may be determined automatically or may be set by a user via input via, for example, the user interface application 44 of
One manner of automatically setting the set point target vector filter time constant is to define a set point target vector filter time factor which defines the operation of the set point target vector filter in relation to the time to steady state of the process. For example, a set point target vector filter time factor could range from 0 to 4 and then the set point target vector filter time constant (defining the time constant of the set point target vector filter) can be determined as the product of the process time to steady state Tss and the set point target vector filter time factor. As will be understood, this set point target vector filter time factor could change with each new process model identification. Of course, the set point target vector filter factor or the set point target vector filter time constant could be selected or provided by the user, if so desired.
Generally speaking, the penalty on error (which defines the coefficient that the MPC controller algorithm applies to an error vector between the desired CV and the predicted CV during the calculation of the controller moves) may be set to one and need not be changeable. In fact, in a single-loop system, only the ratio between the penalty on move (POM) and the penalty on error (PE) matters, meaning that as long as one of these variables is changeable, the other need not be.
In one embodiment, a default penalty on move (which defines the penalty that the MPC control algorithm accesses to unit control signal moves during the calculation of the controller moves) may be calculated every time the process model changes as:
Here DT is the process dead time (from the process model), PH is the prediction horizon, G is the process gain (from the process model) and PM is the calculated penalty on move. Generally speaking, this equation, which has been determined heuristically based on observations, accesses a higher penalty on moves as the process dead time increases, and to a lesser extent, as the process gain increases. Of course, other equations or methods of calculating a recommended or default value for the penalty on move could be used instead.
In any event, the recommended value of the penalty on move may be displayed on the user interface (e.g., using the interface application 44 of
Referring again to
Generally, the block 118 develops an MPC unconstrained incremental controller from the step response model (or step response vector), the penalty on move (POM) and the penalty on error (which is assumed for this discussion to be one). The general solution for the incremental controller with control horizon m and a prediction horizon p at controller scan k is:
ΔU(k)=KE p(k)=(S uTΓyTΓy S u+ΓuTΓu)−1 S uTΓyTΓy E p(k)
It is reasonable to assume that, and the adaptive controller is set up so that, γ=1. As a result the controller gain matrix can be expressed as:
Thus, its easy to see that the single-loop MPC controller with a control horizon of one is simply a process step response vector scaled by the sum of two coefficients:
As a result, it is a simple computational step to regenerate the controller gain matrix K as a closed form, non-recursive and non-matrix equation, to be used in the MPC algorithm block 85 of
In the case of an MPC controller with a control horizon equal to two, the dynamic matrix (transposed) can be expressed as:
In this case, the controller gain matrix K is derived as:
To express this equation more simply, the following variables can be defined:
Then using these variables, the controller gain matrix K can be expressed as:
which can be re-expressed as:
Now, the MPC controller gain matrix for the first controller move of the k1 (i.e., the first controller more of the two move control horizon) is:
As will be understood, therefore, the first move of the two move controller matrix is a resealed step response vector, which to some degree is similar to the control vector for the control horizon equal one. The constant scaling factor for all step response coefficients is:
This coefficient decreases when the penalty on move (u) increases and so an individual scaling factor for every step response coefficient can be expressed as:
This derived formula illustrates, in a clear way, the resemblance of a step response vector and the controller vector. However, this equation is not suitable for the controller calculations because of the uncontrolled possibility of division by zero in the step coefficients ratios. However, a suitable implementation form can be defined as:
To determine this controller gain for the first move, a calculation sequence could be set as follows:
First, calculate m as:
Second, calculate n as:
Third, calculate l as:
Thereafter, rescale the step response by m, and then rescale the step response by l and shift to the right, i.e., make b1=0. Next, subtract the two resealed step responses mB−lB (shifted) and rescale the vector obtained from this step by
to get the controller vector.
In a similar manner, the MPC controller matrix or gain for the second controller move k2 can be expressed as:
Again, as illustrated above, for the case of a single-loop MPC controller with a control horizon of two, the controller gain equations are not computationally excessive and thus can be determined on-line within the controller while the controller is operating to control the process. Additionally, while the controller gain algorithms for an MPC controller with a control horizon of one and two have been provided herein, it is possible that MPC controllers with larger control horizons of, for example, three, four, five or even higher, may be expressed and determined using mathematical computations that lend themselves to be executed in the process controller device and thereby allow adaptive MPC controller regeneration on-line during actual control of a process.
Referring again to
When the download is complete, the block 122 returns control to the block 104 to continue collecting data. It will of course be understood that the control block 104 can continue to collect process data during the operation of the other blocks of the routine 100. Still further, as indicated above, the set point target vector filter factor or time constant and the penalty on move variable are two tuning factors which can be changed by a user at any time, and may be downloaded to the MPC controller to change the operation of the MPC controller 54 at any time. Thus, the MPC controller may be tuned at various times and need not wait to be tuned until the MPC controller is updated based on a new process model. In other words, the MPC controller 54 may by updated with new set point target vector filter factors and new penalty on move values at any time during the operation of the process.
A section 220 of the display screen 200 illustrates information for the various models associated with the process and the MPC controller to provide the user with a better understanding of what is currently happening within the process or within the adaptive MPC portion of the process. The section 220 of
The MPC feedback model section 224 shows (under the column titled “Calculated”) the MPC model that was used to calculate the MPC algorithm and MPC process model currently be used for the feedback path of the MPC controller. In this case, the process controller is currently operating using an MPC controller that was generated based on a process model that had a process gain of 1.00, a time constant of 20.0 and a dead time of 1.0 However, the section 224 also shows the “Current” value of the most recently determined process model for the process as the model to use for the next adaptation cycle, when implemented. This model is illustrated in the example of
The operations section 226 of the screen 200 illustrates the operational region in which the process is currently operating. It is noted that the process can be defined as having separate operational regions as determined by, for example, values of the controlled variables or any other variables. If desired, different models may be selected and used based on the operational region in which the process is running. In fact, some operational regions may be more adaptable to the use of the adaptive MPC controller described herein. Still further, the operations section 226 indicates that the process model estimation unit is currently collecting data and is in a learn mode.
Importantly, the MPC tuning section 228 allows the user to tune the MPC controller by allowing the user to change the SP filter settings and the penalty on move variable. In particular, the user may specify, in this case at an input block 235, an SP filter factor (currently set at 2) which is to be multiplied by the process model time to steady state to determine the SP filter time constant used in the MPC controller. Additionally, a slider bar 238 allows the user to change the penalty on move variable to tune the MPC controller to thereby have a slower or a faster response characteristic. Generally speaking, the slider bar 238 may be used to specify or change a penalty on move factor which is multiplied by the default penalty on move calculated as described above with respect to the block 116 of
In any event, the user display screen 200 can be used by, for example, a control operator control or other user to change the tuning of the MPC controller, to update or cause a process model to be recalculated and updated based on process data selected in the area 210, to cause the MPC controller model to be updated adaptively with the most recently calculated process model, etc. Thus, the user interface screen 200 provides the user with a high degree or level of input over the adaptive updating of the MPC controller, including being able to select the robustness of the controller as defined by the SP filter factor and the penalty on move factor, to define the data to be used to generate a new process model and whether and when to update the MPC controller based on a specific process model. Of course, the adaptive MPC controller block can be completely automatic or semi-automatic in which these functions are performed automatically, periodically or at significant times, such as after a new process model has been determined. Still further, as described above, the adaptive MPC controller may perform automatic calculation of the tuning parameters based on the estimated time to steady state determined by a process model calculation, which eliminates the need to perform initial tuning, a common disadvantage of known MPC controllers. In one embodiment, the initial process model may be determined from process data and the time to steady state may be calculated therefrom, or the user may initially enter the time to steady state which may be used as described above to determine the default tuning parameters for the POM and the SP filter factor, as well as to determine the prediction horizon and the execution time for the controller. Likewise, if desired, the user may specify an initial process model to use when first setting up or running the MPC controller block.
While the adaptive MPC function block has been described and illustrated herein as having an adaptive MPC model generator located within the same function block and therefore executed in the same device as the MPC controller block, it is also possible to implement the adaptive model generator in a separate device, such as in a user interface device. In particular, the adaptive model generator may be located in a different device, such as in one of the user workstations 13 and may communicate with the MPC controller as described in conjunction with
Additionally, while the adaptive MPC controller block and other blocks and routines described herein have been described herein as being used in conjunction with Fieldbus and standard 4-20 ma devices, they can, of course, be implemented using any other process control communication protocol or programming environment and may be used with any other types of devices, function blocks or controllers. Although the adaptive MPC or other adaptive DMC control blocks and the associated generation and viewing routines described herein are preferably implemented in software, they may be implemented in hardware, firmware, etc., and may be executed by any other processor associated with a process control system. Thus, the routine 100 described herein or any portion thereof may be implemented in one or more standard multi-purpose CPUs or on specifically designed hardware or firmware such as, for example, ASICs, if so desired. When implemented in software, the software may be stored in any computer readable memory such as on a magnetic disk, a laser disk, an optical disk, a flash memory or other storage medium, in a RAM or ROM of a computer or processor, etc. Likewise, this software may be delivered to a user or to a process control system via any known or desired delivery method including, for example, on a computer readable disk or other transportable computer storage mechanism or may be modulated and sent over a communication channel such as a telephone line, the internet, etc. (which is viewed as being the same as or interchangeable with providing such software via a transportable storage medium).
Thus, while the present invention has been described with reference to specific examples, which are intended to be illustrative only and not to be limiting of the invention, it will be apparent to those of ordinary skill in the art that changes, additions or deletions may be made to the disclosed embodiments without departing from the spirit and scope of the invention.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US4054408||30 Aug 1976||18 Oct 1977||Shell Oil Company||Method for optimizing the position of a furnace damper without flue gas analyzers|
|US4054780||1 Nov 1976||18 Oct 1977||Phillips Petroleum Company||Gain-adaptive process control|
|US4349869||1 Oct 1979||14 Sep 1982||Shell Oil Company||Dynamic matrix control method|
|US4358822||30 Mar 1981||9 Nov 1982||Sanchez Juan M||Adaptive-predictive control system|
|US4616308||2 Dec 1985||7 Oct 1986||Shell Oil Company||Dynamic process control|
|US4641235||25 Sep 1984||3 Feb 1987||Kabushiki Kaisha Toshiba||Process control apparatus with process dependent switching between process control modes|
|US4736316||6 Aug 1986||5 Apr 1988||Chevron Research Company||Minimum time, optimizing and stabilizing multivariable control method and system using a constraint associated control code|
|US4912624||30 Mar 1988||27 Mar 1990||Syracuse University||Multi-parameter optimization circuit|
|US5018215||23 Mar 1990||21 May 1991||Honeywell Inc.||Knowledge and model based adaptive signal processor|
|US5043863||2 May 1989||27 Aug 1991||The Foxboro Company||Multivariable adaptive feedforward controller|
|US5159547||16 Oct 1990||27 Oct 1992||Rockwell International Corporation||Self-monitoring tuner for feedback controller|
|US5159562||26 Mar 1992||27 Oct 1992||Westinghouse Electric Corp.||Optimization of a plurality of multiple-fuel fired boilers using iterated linear programming|
|US5180896||11 Oct 1990||19 Jan 1993||University Of Florida||System and method for in-line heating of medical fluid|
|US5216593||24 Jan 1991||1 Jun 1993||International Business Machines Corporation||Method and apparatus for discrete activity resourse allocation through cardinality constraint generation|
|US5268835||19 Sep 1991||7 Dec 1993||Hitachi, Ltd.||Process controller for controlling a process to a target state|
|US5272621||9 Nov 1989||21 Dec 1993||Nippon Denki Garasu Kabushiki Kaisha||Method and apparatus using fuzzy logic for controlling a process having dead time|
|US5283729||30 Aug 1991||1 Feb 1994||Fisher-Rosemount Systems, Inc.||Tuning arrangement for turning the control parameters of a controller|
|US5301101||14 Apr 1993||5 Apr 1994||Honeywell Inc.||Receding horizon based adaptive control having means for minimizing operating costs|
|US5329443||16 Jun 1992||12 Jul 1994||Praxair Technology, Inc.||Two-phase method for real time process control|
|US5335164||31 Jul 1991||2 Aug 1994||Universal Dynamics Limited||Method and apparatus for adaptive control|
|US5347446||10 Feb 1992||13 Sep 1994||Kabushiki Kaisha Toshiba||Model predictive control apparatus|
|US5351184||26 Jan 1993||27 Sep 1994||Honeywell Inc.||Method of multivariable predictive control utilizing range control|
|US5353207||10 Jun 1992||4 Oct 1994||Pavilion Technologies, Inc.||Residual activation neural network|
|US5384698||31 Aug 1992||24 Jan 1995||Honeywell Inc.||Structured multiple-input multiple-output rate-optimal controller|
|US5386373||5 Aug 1993||31 Jan 1995||Pavilion Technologies, Inc.||Virtual continuous emission monitoring system with sensor validation|
|US5408406||7 Oct 1993||18 Apr 1995||Honeywell Inc.||Neural net based disturbance predictor for model predictive control|
|US5453925||28 May 1993||26 Sep 1995||Fisher Controls International, Inc.||System and method for automatically tuning a process controller|
|US5457625||13 Apr 1994||10 Oct 1995||The M. W. Kellogg Company||Maximizing process production rates using permanent constraints|
|US5461559||4 Oct 1993||24 Oct 1995||The United States Of America As Represented By The Secretary Of The Air Force||Hierarchical control system for molecular beam epitaxy|
|US5477449||14 Apr 1994||19 Dec 1995||Kabushiki Kaisha Toshiba||Hierarchical model predictive control system based on accurate manipulation variable prediction|
|US5517424||31 Mar 1994||14 May 1996||Electric Power Research Institute, Inc.||Steam turbine fuzzy logic cyclic control method and apparatus therefor|
|US5521814||18 Oct 1994||28 May 1996||Betz Laboratories, Inc.||Process optimization and control system that plots inter-relationships between variables to meet an objective|
|US5537310 *||20 Dec 1994||16 Jul 1996||Yamatake-Honeywell Co., Ltd.||Internal Model Controller with automatically correcting gain of the internal model controller|
|US5539638||5 Nov 1993||23 Jul 1996||Pavilion Technologies, Inc.||Virtual emissions monitor for automobile|
|US5548528||30 Jan 1995||20 Aug 1996||Pavilion Technologies||Virtual continuous emission monitoring system|
|US5559690||16 Sep 1994||24 Sep 1996||Pavilion Technologies, Inc.||Residual activation neural network|
|US5561599||14 Jun 1995||1 Oct 1996||Honeywell Inc.||Method of incorporating independent feedforward control in a multivariable predictive controller|
|US5568378||24 Oct 1994||22 Oct 1996||Fisher-Rosemount Systems, Inc.||Variable horizon predictor for controlling dead time dominant processes, multivariable interactive processes, and processes with time variant dynamics|
|US5572420||3 Apr 1995||5 Nov 1996||Honeywell Inc.||Method of optimal controller design for multivariable predictive control utilizing range control|
|US5574638||3 Apr 1995||12 Nov 1996||Lu; Zhuxin J.||Method of optimal scaling of variables in a multivariable predictive controller utilizing range control|
|US5586221||1 Jul 1994||17 Dec 1996||Syracuse University||Predictive control of rolling mills using neural network gauge estimation|
|US5587897||27 Dec 1994||24 Dec 1996||Nec Corporation||Optimization device|
|US5587899||10 Jun 1994||24 Dec 1996||Fisher-Rosemount Systems, Inc.||Method and apparatus for determining the ultimate gain and ultimate period of a controlled process|
|US5594858||29 Jul 1993||14 Jan 1997||Fisher-Rosemount Systems, Inc.||Uniform control template generating system and method for process control programming|
|US5625552||13 Jul 1995||29 Apr 1997||Mathur; Anoop K.||Closed loop neural network automatic tuner|
|US5659467||26 Jun 1996||19 Aug 1997||Texas Instruments Incorporated||Multiple model supervisor control system and method of operation|
|US5659667||17 Jan 1995||19 Aug 1997||The Regents Of The University Of California Office Of Technology Transfer||Adaptive model predictive process control using neural networks|
|US5673368||10 Nov 1994||30 Sep 1997||Siemens Aktiengesellschaft||Method and device for conducting a process in a controlled system with at least one precomputed process parameter determined using a mathematical model having variable model parameters adjusted based on a network response of a neural network|
|US5680304||21 Aug 1996||21 Oct 1997||Motorola, Inc.||Method for controlling nonlinear systems|
|US5680409||11 Aug 1995||21 Oct 1997||Fisher-Rosemount Systems, Inc.||Method and apparatus for detecting and identifying faulty sensors in a process|
|US5682309||28 Apr 1995||28 Oct 1997||Exxon Chemical Patents Inc.||Feedback method for controlling non-linear processes|
|US5682317||23 Jul 1996||28 Oct 1997||Pavilion Technologies, Inc.||Virtual emissions monitor for automobile and associated control system|
|US5696696||11 Oct 1994||9 Dec 1997||Betzdearborn, Inc.||Apparatus and method for automatically achieving and maintaining congruent control in an industrial boiler|
|US5704011||1 Nov 1994||30 Dec 1997||The Foxboro Company||Method and apparatus for providing multivariable nonlinear control|
|US5727128||8 May 1996||10 Mar 1998||Fisher-Rosemount Systems, Inc.||System and method for automatically determining a set of variables for use in creating a process model|
|US5740033||13 Oct 1992||14 Apr 1998||The Dow Chemical Company||Model predictive controller|
|US5748467||21 Feb 1995||5 May 1998||Fisher-Rosemont Systems, Inc.||Method of adapting and applying control parameters in non-linear process controllers|
|US5758047||14 Jun 1995||26 May 1998||Lu; Zhuxin Joseph||Method of process controller optimization in a multivariable predictive controller|
|US5781432||4 Dec 1996||14 Jul 1998||Pavilion Technologies, Inc.||Method and apparatus for analyzing a neural network within desired operating parameter constraints|
|US5825646||3 Jun 1996||20 Oct 1998||Pavilion Technologies, Inc.||Method and apparatus for determining the sensitivity of inputs to a neural network on output parameters|
|US5838563||12 Apr 1996||17 Nov 1998||Fisher-Rosemont Systems, Inc.||System for configuring a process control environment|
|US5838595||25 Nov 1996||17 Nov 1998||Texas Instruments, Inc.||Apparatus and method for model based process control|
|US5841652||3 Apr 1996||24 Nov 1998||Scap Europa, S.A.||Adaptive-predictive control and optimization system|
|US5842189||24 Sep 1997||24 Nov 1998||Pavilion Technologies, Inc.||Method for operating a neural network with missing and/or incomplete data|
|US5859773||23 Sep 1996||12 Jan 1999||Pavilion Technologies, Inc.||Residual activation neural network|
|US5873251||13 Sep 1996||23 Feb 1999||Kabushiki Kaisha Toshiba||Plant operation control system|
|US5877954||3 May 1996||2 Mar 1999||Aspen Technology, Inc.||Hybrid linear-neural network process control|
|US5886895||1 Jul 1996||23 Mar 1999||Kabushiki Kaisha Toshiba||Plant utility optimizing method and an optimizing system|
|US5920478||27 Jun 1997||6 Jul 1999||Oakleaf Engineering, Inc.||Multi-input multi-output generic non-interacting controller|
|US5933345||6 May 1996||3 Aug 1999||Pavilion Technologies, Inc.||Method and apparatus for dynamic and steady state modeling over a desired path between two end points|
|US5970430||3 Sep 1997||19 Oct 1999||Fisher Controls International, Inc.||Local device and process diagnostics in a process control network having distributed control functions|
|US6041320||23 Aug 1993||21 Mar 2000||Fisher Controls International, Inc.||Multi-region fuzzy logic control system with auxiliary variables|
|US6047221||3 Oct 1997||4 Apr 2000||Pavilion Technologies, Inc.||Method for steady-state identification based upon identified dynamics|
|US6049738||3 Mar 1997||11 Apr 2000||Hitachi, Ltd.||Control model modeling support system and a method therefor|
|US6055483||5 May 1997||25 Apr 2000||Honeywell, Inc.||Systems and methods using bridge models to globally optimize a process facility|
|US6056781||6 Feb 1998||2 May 2000||The Dow Chemical Company||Model predictive controller|
|US6064809||5 Jun 1998||16 May 2000||The Board Of Trustees Of The University Of Illinois||Fast model predictive ellipsoid control process|
|US6079205 *||15 Sep 1998||27 Jun 2000||Honda Giken Kogyo Kabushiki Kaisha||Plant control system|
|US6122555||5 May 1997||19 Sep 2000||Honeywell International Inc.||System and methods for globally optimizing a process facility|
|US6207936||30 Jan 1997||27 Mar 2001||Asm America, Inc.||Model-based predictive control of thermal processing|
|US6215907||26 Jun 1998||10 Apr 2001||Fisher-Rosemont Systems, Inc.||Recursive on-line wavelet data compression technique for use in data storage and communications|
|US6249712||25 Sep 1996||19 Jun 2001||William J. N-O. Boiquaye||Adaptive control process and system|
|US6253113||20 Aug 1998||26 Jun 2001||Honeywell International Inc||Controllers that determine optimal tuning parameters for use in process control systems and methods of operating the same|
|US6266605 *||17 Dec 1999||24 Jul 2001||Honda Giken Kogyo Kabushiki||Plant control system|
|US6269274||9 Sep 1998||31 Jul 2001||Honeywell International Inc||System and method for balancing and distributing control algorithm processing load and real time plant control system employing the same|
|US6278899||6 Oct 1998||21 Aug 2001||Pavilion Technologies, Inc.||Method for on-line optimization of a plant|
|US6298454||22 Feb 1999||2 Oct 2001||Fisher-Rosemount Systems, Inc.||Diagnostics in a process control system|
|US6319362||17 Nov 1998||20 Nov 2001||Metso Paper Automation Oy||Method and equipment for controlling properties of paper|
|US6328851||17 Nov 1998||11 Dec 2001||Metso Paper Automation Oy||Method and equipment for controlling properties of paper|
|US6330483||7 May 1999||11 Dec 2001||The Boeing Company||Optimal control system|
|US6336050||3 Feb 1998||1 Jan 2002||British Telecommunications Public Limited Company||Method and apparatus for iteratively optimizing functional outputs with respect to inputs|
|US6360131||16 Oct 1998||19 Mar 2002||Cheng||Model-free adaptive control for flexible production systems|
|US6373033||27 Jun 2000||16 Apr 2002||Asm America, Inc.||Model-based predictive control of thermal processing|
|US6381505||11 Mar 1999||30 Apr 2002||Aspen Technology, Inc.||Robust steady-state target calculation for model predictive control|
|US6438430||9 May 2000||20 Aug 2002||Pavilion Technologies, Inc.||Kiln thermal and combustion control|
|US6445963||4 Oct 1999||3 Sep 2002||Fisher Rosemount Systems, Inc.||Integrated advanced control blocks in process control systems|
|US6453308||24 Sep 1998||17 Sep 2002||Aspen Technology, Inc.||Non-linear dynamic predictive device|
|US6484133||31 Mar 2000||19 Nov 2002||The University Of Chicago||Sensor response rate accelerator|
|US6487459||16 Feb 1999||26 Nov 2002||Pavilion Technologies, Inc.||Method and apparatus for modeling dynamic and steady-state processes for prediction, control and optimization|
|US6493596||28 Feb 2000||10 Dec 2002||Pavilion Technologies, Inc.||Method and apparatus for controlling a non-linear mill|
|US6493601||18 Jun 1999||10 Dec 2002||Impact Systems, Inc.||Real time measurement system for a moving web using a Kalman filter algorithm|
|US6542782||31 Dec 1998||1 Apr 2003||Z. Joseph Lu||Systems for generating and using a lookup table with process facility control systems and models of the same, and methods of operating such systems|
|US6577908||20 Jun 2000||10 Jun 2003||Fisher Rosemount Systems, Inc||Adaptive feedback/feedforward PID controller|
|US6577916||10 Mar 2000||10 Jun 2003||Voith Sulzer Papiertechnik Patent Gmbh||Method of carrying out an optimized fiber or paper manufacturing process|
|US6594530||4 Jun 2002||15 Jul 2003||Fieldbus Foundation||Block-oriented control system|
|US6611737||20 Apr 2000||26 Aug 2003||Canadian Space Agency||Advanced ship autopilot system|
|US6615090||7 Feb 2000||2 Sep 2003||Fisher-Rosemont Systems, Inc.||Diagnostics in a process control system which uses multi-variable control techniques|
|US6625501||14 Aug 2002||23 Sep 2003||Pavilion Technologies, Inc.||Kiln thermal and combustion control|
|US6678668||7 Mar 2002||13 Jan 2004||Insyst Ltd.||System and method for complex process optimization and control|
|US6681155||31 Aug 1999||20 Jan 2004||Mitsubishi Chemical Corporation||Optimizing control method and optimizing control system for power plant|
|US6701223||11 Sep 2000||2 Mar 2004||Advantica, Inc.||Method and apparatus for determining optimal control settings of a pipeline|
|US6714899||28 Mar 2001||30 Mar 2004||Aspen Technology, Inc.||Robust steady-state target calculation for model predictive control|
|US6721609||14 Jun 2000||13 Apr 2004||Fisher-Rosemount Systems, Inc.||Integrated optimal model predictive control in a process control system|
|US6735483||9 Dec 2002||11 May 2004||Pavilion Technologies, Inc.||Method and apparatus for controlling a non-linear mill|
|US6745088 *||11 Jun 2001||1 Jun 2004||The Dow Chemical Company||Multi-variable matrix process control|
|US6772019||14 Nov 2001||3 Aug 2004||Lockheed Martin Corporation||Method and system for multi-parameter choice optimization|
|US6826521||6 Apr 2000||30 Nov 2004||Abb Automation Inc.||System and methodology and adaptive, linear model predictive control based on rigorous, nonlinear process model|
|US6901300||7 Feb 2002||31 May 2005||Fisher-Rosemount Systems, Inc..||Adaptation of advanced process control blocks in response to variable process delay|
|US6954713||10 Jan 2002||11 Oct 2005||Fisher-Rosemount Systems, Inc.||Cavitation detection in a process plant|
|US6993396||20 Mar 2003||31 Jan 2006||John Peter Gerry||System for determining the health of process control feedback loops according to performance assessment criteria|
|US7016779||31 Jan 2003||21 Mar 2006||Cambridge Consultants Limited||Control system|
|US7043409||1 Dec 1999||9 May 2006||Siemens Aktiengesellschaft||Method and device for designing a technical system|
|US7050880 *||30 Dec 2003||23 May 2006||Sc Solutions||Chemical-mechanical planarization controller|
|US7272454||5 Jun 2003||18 Sep 2007||Fisher-Rosemount Systems, Inc.||Multiple-input/multiple-output control blocks with non-linear predictive capabilities|
|US7395124 *||10 Nov 2005||1 Jul 2008||Rockwell Automation Technologies, Inc.||Adaptive motor drive method and apparatus including inertia estimator|
|US7451004||30 Sep 2005||11 Nov 2008||Fisher-Rosemount Systems, Inc.||On-line adaptive model predictive control in a process control system|
|US20020016640||11 Jun 2001||7 Feb 2002||Gagne Ronald A.||Multi-variable matrix process control|
|US20020103548||5 Dec 2000||1 Aug 2002||Steven Treiber||Computer method and apparatus for optimized controller in a non-linear process|
|US20030120361||12 Mar 2001||26 Jun 2003||Anderson Ketil Strand||Process control system|
|US20030139825||18 Dec 2002||24 Jul 2003||Lund Richard A.||Method of ascertaining control parameters for a control system|
|US20030195641||21 Apr 2003||16 Oct 2003||Wojsznis Wilhelm K.||State based adaptive feedback feedforward PID controller|
|US20040049299||11 Sep 2002||11 Mar 2004||Wilhelm Wojsznis||Integrated model predictive control and optimization within a process control system|
|US20040117766||25 Jul 2003||17 Jun 2004||Fisher-Rosemount Systems, Inc.||Integrated model predictive control and optimization within a process control system|
|US20040181300||11 Mar 2003||16 Sep 2004||Clark Robert L.||Methods, apparatus and computer program products for adaptively controlling a system by combining recursive system identification with generalized predictive control|
|US20050209714||4 Feb 2005||22 Sep 2005||Rawlings James B||SISO model predictive controller|
|US20070078529 *||30 Sep 2005||5 Apr 2007||Fisher-Rosemount Systems, Inc.||On-line adaptive model predictive control in a process control system|
|EP0037579A2||7 Apr 1981||14 Oct 1981||Juan Martin Sanchez||Adaptive-predictive control method and adaptive-predictive control system|
|EP0524317A1||10 Feb 1992||27 Jan 1993||Kabushiki Kaisha Toshiba||Model forecasting controller|
|EP0588594A2||14 Sep 1993||23 Mar 1994||TEXACO, Inc.||Control system using an adaptive neural network for target and path optimization for a multivariable, nonlinear process|
|EP0643344A1||19 Jul 1994||15 Mar 1995||Texas Instruments Incorporated||Apparatus and method for model based process control|
|EP1600322A2||17 May 2005||30 Nov 2005||Nissan Motor Company, Limited||Model predictive control apparatus|
|GB1355545A||Title not available|
|GB2186715A||Title not available|
|GB2394312A||Title not available|
|GB2394313A||Title not available|
|GB2430764B||Title not available|
|1||"DeltaV Predict and DeltaV PredictPro-Powerful Practical Model Predictive Control," Product data sheet, Jun. 2003.|
|2||"DeltaV Predict and DeltaV PredictPro—Powerful Practical Model Predictive Control," Product data sheet, Jun. 2003.|
|3||Babb, "Fast Computers Open the Way for Advanced Controls," Control Engineering, Jan. 1991.|
|4||Bitton et al., "Feasibility and Performance Study of Dependency Inference [database design]," Data Engineering, 1989. Proceedings. Fifth International Conference on, Feb. 6-10, 1989 pp. 635-641.|
|5||Brochure entitled "Advanced Control and Optimization, Capturing and Sustaining Maximum Client Benefits" from Aspen Technology, Inc. of Cambridge, MA, copyright 1996.|
|6||Chinese Office Action and English-language translation thereof for Application No. 200610131752.7, dated Sep. 25, 2009.|
|7||Chmelyk, "An Integrated Approach to Model Predictive Control of an Industrial Lime Kiln," NORPAC Controls Ltd.|
|8||Cutler et al., "Dynamic Matrix Control-A Computer Control Algorithm," Proceedings of the Joint Automatic Control Conference, 1980.|
|9||Cutler et al., "Dynamic Matrix Control—A Computer Control Algorithm," Proceedings of the Joint Automatic Control Conference, 1980.|
|10||Dabrowski et al., "Design of Switched-Capacitor FIR Filters with Application to a Low-Power MFSK Receiver," Circuits, Devices and Systems, IEE Proceedings G, vol. 139, Issue: 4, Aug. 1992 pp. 450-466.|
|11||De Keyser et al., "A Comparative Study of Self-Adaptive Long-Range Predictive Control Methods," Automatica, vol. 24, No. 2, pp. 149-163, 1988.|
|12||Dynamic Matrix Corporation, "DMC Multivariable Model Predictive Constraint Control."|
|13||Edgar et al., "Optimization of Chemical Processes," Chapter 7, McGraw-Hill, Inc., pp. 251-298 (1988).|
|14||Forrest et al., IBM System Journal, Mar. 1992, "Implementing Interior Point Linear Programing Methods in the Optimization Subroutine Library-Technical."|
|15||Forrest et al., IBM System Journal, Mar. 1992, "Implementing Interior Point Linear Programing Methods in the Optimization Subroutine Library—Technical."|
|16||Froisy, "Model Predictive Control: Past, Present and Future," 33 ISA Transactions pp. 235-243 (1994).|
|17||Garcia et al., "Model Predictive Control: Theory and Practice-a Survey," Automatica, vol. 25, No. 3, pp. 335-348, 1989.|
|18||Garcia et al., "Model Predictive Control: Theory and Practice—a Survey," Automatica, vol. 25, No. 3, pp. 335-348, 1989.|
|19||Garcia et al., "Quadratic Programming Solution of Dynamic Matrix Control (QDMC)," Chemical Engineering Communications, 46:73-87, 1986.|
|20||Gschwendtner et al., "Adaptive Propagation Modeling Based on Neural Network Techniques," Fellow, IEEE, 1996 IEEE.|
|21||Hanagud et al., "Artificial Intelligence-Based Model-Adaptive Approach to Flexible Structure Control," American Institute of Aeronautics and Astronautics, Inc., vol. 13, pp. 534-544 (1990).|
|22||Hoffman et al., "Model Predictive Control: Future Requirements and Trends," ISA, 1993, Paper #93-369, pp. 1241-1247 (Oct. 1993).|
|23||Honeywell Hi-Spec Solutions: "Honeywell's RMPCT-A Breakthrough in Process Control Technology" Frequently Asked Questions.|
|24||Honeywell Hi-Spec Solutions: "Honeywell's RMPCT—A Breakthrough in Process Control Technology" Frequently Asked Questions.|
|25||Huang et al., "100 Tap FIR/IIR Analog Linear-Phase Lowpass Filter Qiuting," VLSI Circuits, 1995. Digest of Technical Papers., 1995 Symposium on, Jun. 8-10, 1995 pp. 91-92.|
|26||Huang, "Mixed Analog/Digital, FIR/IIR Realization of a Linear-Phase Lowpass Filter," Solid-State Circuits, IEEE Journal of, vol. 31, Issue: 9, Sep. 1996 pp. 1347-1350.|
|27||Lee et al., "State-space Interpretation of Model Predictive Control," Automatica, vol. 30, No. 4, pp. 707-717, 1994.|
|28||Lee et al., "Tuning of Model Predictive Controllers for Robust Performance," Computers in Chemical Engineering, vol. 18, No. 1, pp. 15-37, 1994.|
|29||Lynch at al., "An Object Oriented Intelligent Control Architecture," 1992 ACC/TP4, vol. 3, 92CH3072-6, Chicago.|
|30||MacArthur, "RMPCT: A New Robust Approach to Multivariable Predictive Control for the Process Industries," Proceedings of the 1996 Control Systems Conference, Halifax, Canada, 1996.|
|31||Maciejowski, "Model Predictive Control with Constraints," Pearson Education Limited, Appendix A, Appendix B, 2002.|
|32||McDonnell et al., "Predictive Functional Control of Multivariable Systems with more Outputs than Inputs," Proceedings 28 CDC, Orlando, Florida, 1989.|
|33||McMillian, "Process Control Optimization," slides entitled "Breaking Into Model Based Control" presented at Fisher-Rosemount Systems Advanced Control Seminar, San Antonio (Nov. 1997).|
|34||MDC Technology, "RTO Technical Presentation."|
|35||Microelectronics (Digital and Analog Circuits and Systems), Jacob Millman, Ph.D., Copyright © 1979 McGraw-Hill Inc.|
|36||Misra, "Auxiliary Variables Notes on UNITY" 15-90, UT Austin (1990).|
|37||Morari, et al., "Model Predictive Control Toolbox," Matlab® User's Guide, Version 1, The Math Works, Inc. 1995.|
|38||Qin et al., "An Overview of Industrial Model Predictive Control Technology," AIChE Conference, pp. 1-31 (1996).|
|39||Rao et al., "Application of Interior-Point Methods to Model Predictive Control," J. Optim. Theory Appl., 99:723-757, 1998.|
|40||Rawlings et al., "Model Predictive Control with Linear Models," Process Systems Engineering, vol. 39, No. 2, Feb. 1993.|
|41||Rumbaugh, "Let There Be Objects: A Short Guide to Reification," Journal of Object-Oriented Programming, vol. 5, No. 7, Nov./Dec. 1992.|
|42||Search Report for Application No. GB0619184.5, dated Jan. 23, 2007.|
|43||Shridhar et al., "A Novel Tuning Strategy for Multivariable Model Predictive Control," ISA Transactions, vol. 36, No. 4, pp. 273-280, 1998.|
|44||Thiele et al., "Configuration and Viewing Display for an Integrated Model Predictive Control and Optimizer Function Block," U.S. Appl. No. 10/310,416, filed Dec. 5, 2002.|
|45||Thiele et al., "Simulation of Multivariable Model Predictive Control," in Proceedings of ISA 2001 Conference, Sep. 10-13, 2001, Houston, TX.|
|46||Thiele, "Benefits and challenges of implementing model predictive control as a function block," Advances in Instrumentation and Control, ISA/2000 Conference, Aug. 2000, New Orleans.|
|47||U.S. Appl. No. 09/412,078, dated Oct. 4, 1999, "Integrated Advanced Control Blocks in Process Control Systems."|
|48||U.S. Appl. No. 60/132,780, dated May 6, 1999, "Integrating Distributed Process Control System Functionality on a Single Computer."|
|49||Wilhelm et al., "Advanced Process Control in Record Time," Control Magazine Software and Information Systems (May 1999).|
|50||Wilkins et al., "The Use of Adaptive Algorithms for Obtaining Optimal Electrical Shimming in Magnetic Resonance Imaging (MRI)," IEEE Transactions on Biomedical Engineering, vol. 36, No. 2, Feb. 1989, pp. 202-210.|
|51||Wojsznis et al., "Easy Robust Optimal Predictive Controller," Advances in Instrumentation and Control, ISA/2000 Conference, Aug. 2000, New Orleans.|
|52||Wojsznis et al., "Practical Approach to Tuning MPC," in Proceedings of ISA 2001 Conference, Sep. 10-13, 2001, Houston, TX.|
|53||Wojsznis et al., "Robust Predictive Controller in Object-Oriented Implementation," ISA, Paper No. 93-297, pp. 521-528, 1993.|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US8155764 *||11 Feb 2010||10 Apr 2012||Honeywell International Inc.||Multivariable model predictive control for coalbed gas production|
|US8185217 *||30 Jan 2009||22 May 2012||Fisher-Rosemount Systems, Inc.||Robust adaptive model predictive controller with tuning to compensate for model mismatch|
|US8280533 *||22 Jun 2009||2 Oct 2012||Fisher-Rosemount Systems, Inc.||Continuously scheduled model parameter based adaptive controller|
|US8463477 *||4 Aug 2011||11 Jun 2013||Robert Bosch Gmbh||Method and system for controlling an electrical machine in a drivetrain of a motor vehicle|
|US9163828||31 Oct 2011||20 Oct 2015||Emerson Process Management Power & Water Solutions, Inc.||Model-based load demand control|
|US9217565||7 Feb 2011||22 Dec 2015||Emerson Process Management Power & Water Solutions, Inc.||Dynamic matrix control of steam temperature with prevention of saturated steam entry into superheater|
|US9235657||13 Mar 2013||12 Jan 2016||Johnson Controls Technology Company||System identification and model development|
|US9335042||16 Aug 2010||10 May 2016||Emerson Process Management Power & Water Solutions, Inc.||Steam temperature control using dynamic matrix control|
|US9436179||13 Mar 2013||6 Sep 2016||Johnson Controls Technology Company||Systems and methods for energy cost optimization in a building system|
|US9447963||7 Feb 2011||20 Sep 2016||Emerson Process Management Power & Water Solutions, Inc.||Dynamic tuning of dynamic matrix control of steam temperature|
|US9488974 *||17 Aug 2010||8 Nov 2016||Massachusetts Institute Of Technology||Method and apparatus for pulse-modulated feedback control|
|US9581980 *||18 Oct 2012||28 Feb 2017||Abb Schweiz Ag||Method and system for updating a model in a model predictive controller|
|US20090198350 *||30 Jan 2009||6 Aug 2009||Fisher-Rosemount Systems, Inc||Robust adaptive model predictive controller with tuning to compensate for model mismatch|
|US20090319060 *||22 Jun 2009||24 Dec 2009||Fisher-Rosemount Systems, Inc.||Continuously Scheduled Model Parameter Based Adaptive Controller|
|US20100222911 *||11 Feb 2010||2 Sep 2010||Honeywell International Inc.||Multivariable model predictive control for coalbed gas production|
|US20120046793 *||17 Aug 2010||23 Feb 2012||Massachusetts Institute Of Technology||Method and apparatus for pulse-modulated feedback control|
|US20120053770 *||4 Aug 2011||1 Mar 2012||Markus Kretschmer||Method and system for controlling an electrical machine in a drivetrain of a motor vehicle|
|US20130041482 *||18 Oct 2012||14 Feb 2013||Abb Research Ltd||Method and system for updating a model in a model predictive controller|
|US20150005945 *||27 Jun 2013||1 Jan 2015||The Boeing Company||Real-Time Feedback Control for Performing Tooling Operations in Assembly Processes|
|EP2314318A1||31 Jan 2002||27 Apr 2011||Biogen Idec Inc.||CD80 antibody for use in combination with chemotherapeutics to treat B cell malignancies|
|U.S. Classification||700/32, 700/29, 700/44, 700/37, 700/28, 700/45, 700/31, 700/30|
|29 Apr 2009||AS||Assignment|
Owner name: FISHER-ROSEMOUNT SYSTEMS, INC., TEXAS
Free format text: CORRECTIVE ASSIGNMENT TO CORRECT THE SERIAL NUMBER PREVIOUSLY RECORDED ON REEL 022590 FRAME 0752;ASSIGNORS:THIELE, DIRK;WOJSZINS, WILHELM K.;REEL/FRAME:022614/0677
Effective date: 20060327
Owner name: FISHER-ROSEMOUNT SYSTEMS, INC., TEXAS
Free format text: CORRECTIVE ASSIGNMENT TO CORRECT THE SERIAL NUMBER PREVIOUSLY RECORDED ON REEL 022590 FRAME 0752. ASSIGNOR(S) HEREBY CONFIRMS THE APPLICATION NO. 11/240,705 SHOULD BE DELETED;ASSIGNORS:THIELE, DIRK;WOJSZINS, WILHELM K.;REEL/FRAME:022614/0677
Effective date: 20060327
|13 May 2009||AS||Assignment|
Owner name: FISHER-ROSEMOUNT SYSTEMS, INC., TEXAS
Free format text: CORRECTIVE ASSIGNMENT TO CORRECT THE ASSINGOR S NAME PREVIOUSLY RECORDED ON REEL 022614 FRAME 0677;ASSIGNORS:THIELE, DIRK;WOJSZNIS, WILHELM K.;REEL/FRAME:022679/0880
Effective date: 20060327
Owner name: FISHER-ROSEMOUNT SYSTEMS, INC., TEXAS
Free format text: CORRECTIVE ASSIGNMENT TO CORRECT THE ASSINGOR S NAME PREVIOUSLY RECORDED ON REEL 022614 FRAME 0677. ASSIGNOR(S) HEREBY CONFIRMS THE CORRECT SPELLING OF ASSIGNOR S NAME IS WILHELM K. WOJSZNIS;ASSIGNORS:THIELE, DIRK;WOJSZNIS, WILHELM K.;REEL/FRAME:022679/0880
Effective date: 20060327
|9 Aug 2011||CC||Certificate of correction|
|23 Jun 2014||FPAY||Fee payment|
Year of fee payment: 4